Miquido

Traditional cloud security relies on perimeter firewalls and strict server-side access controls. Edge AI security, by contrast, operates in an environment where physical devices, binaries, and local memory remain completely exposed to end-user extraction and reverse engineering. Learn more in today's article 👇

Every request to a cloud AI model is a real cost. Multiply it by millions of users, and a flagship feature can quietly start eating your margins. Where your AI runs is a unit economics decision. On-device processing has zero marginal cost per query, and it also brings lower latency, offline capability and an easier path to GDPR compliance. The setup that works best is hybrid: simple, frequent and privacy-sensitive tasks on the phone, and the cloud only where you truly need complex reasoning. In our latest article, we show how to decide which AI features belong on the device, and what trade-offs to expect around monitoring and analytics. If you’re wondering how much your AI features cost in the cloud and what could move to the edge, our engineering team is happy to take a look at your architecture.

Our Head of Technology, Piotrek Polus, will be taking the stage tomorrow alongside PLAY and Scaleway at the AIBA conference!

A camera at the self-checkout that recognises your apples. An in-app "GPS" that walks you to the right shelf. A bank's own map in Fortnite. Each brand started from a moment people already had (creating, paying, browsing an aisle, sharing, playing) and made it easier or more personal. See how LPP, Lidl, Rossmann, Spotify and PKO BP did it. Which step in your customers' mobile journey would you enhance first? We're collecting cases like these for a bigger report on Mobile 2030. Follow Miquido to read it when it's published.

Your low-code prototype worked. That’s when the harder technology decision begins. As the product gains users, integrations and business importance, the architecture that helped you validate quickly may start limiting how safely you can scale. So when should low-code stay? When does custom architecture make more sense? And how is AI changing that decision? We answer 10 questions CTOs should ask before a prototype becomes a platform. Because the goal is to know when the architecture needs to evolve. Read the full article ↓

10 million downloads. A 5.0 rating on the App Store. No. 4 most popular utilities app in Poland. When PLAY needed to move more of its sales and account management out of physical stores and into one mobile app, speed and reliability couldn’t be an afterthought. We built it natively in Swift and Kotlin, wired in in-app payments and instant top-ups, and kept the codebase clean enough to ship new features as fast as Play’s users ask for them. 🟣 Native iOS (Swift) and Android (Kotlin) 🟣 In-app payments and instant top-ups, no branch visit needed 🟣 Automated privacy checks built into every workflow The result: an app that now serves millions of active users, every single day. Engineered for speed. Designed to scale. Watch how it works.

49% of Gen Alpha trust influencers as much as they trust their family and friends. Gen Z and Gen Alpha get lumped into “young people”, but they use products in very different ways. One grew up with the phone as just another tool. The other learns from 30-second tutorials, and your six-screen welcome flow was not built for them. Swipe for what changes in onboarding and account design. Spoiler: the welcome screens are the first thing to go. Which step in your onboarding would a 30-second tutorial generation skip first?

Inline autocomplete made developers a bit faster. It did not fix the bottlenecks that cost enterprise teams the most: multi-file refactoring, legacy code, and debugging that eats whole sprints. Agentic engineering works as a closed loop: the agent plans the change, runs tests in an isolated sandbox, reads linter errors, and corrects itself before a human reviews anything. Engineers move from writing syntax to acting as system architects. Read the full article.

Outsourcing mobile development should give you more capacity. It should not mean giving up control of your product. The real questions start after the contract is signed: - Who owns the architecture? - Who controls the source code and infrastructure? - How much visibility do you have into delivery? - And could another team take over if needed? After years of building and scaling mobile products at Miquido, one principle keeps coming back: Outsource execution. Keep ownership. In our latest LinkedIn article, we share what we’ve learned about building outsourcing models that add engineering capacity without creating dependency. Read the full article below.

78 million BNPL transactions in Poland in 2025. 3.3 million customers. More than half of those transactions were under PLN 100. Few people will fill in a form for a purchase that small. The financing has to fit into the moment of buying. So how does a mobile-first BNPL experience work, from need to payment? We broke it down into six steps: Need arises → Solution placement → Mobile execution → Fast approval → Payment in checkout → Repayment management Financing appears where the decision is made. There are no branch visits and no document exchange. The customer sees the amount, the schedule and the total cost before they decide, and again after approval. When the terms are clear and the costs are visible upfront, BNPL feels like part of checkout. From 20 November 2026, CCD2 puts the UX of credit under the law. Swipe through the carousel for the full journey. Then check one thing in your own product: at which of these six steps do you lose the most customers?